OpenCL and Fuji
I notice in this https://www.phaseone.com/en/Search/Arti ... cleid=1720 that it says OpenCL will be disabled when processing xtrans files - by this I assume it means Fuji RAW files?
Is this correct?
Is this a major slowdown with C1 with OpenCL and Fuji?
Please tell me this is not so ...
Michael
Is this correct?
Is this a major slowdown with C1 with OpenCL and Fuji?
Please tell me this is not so ...
Michael
0
-
You would be best to raise a technical support request ... I don't see any difference except I cannot export as DNG. 0 -
[quote="mtissington" wrote:
I notice in this https://www.phaseone.com/en/Search/Arti ... cleid=1720 that it says OpenCL will be disabled when processing xtrans files - by this I assume it means Fuji RAW files?
Is this correct?
Is this a major slowdown with C1 with OpenCL and Fuji?
Please tell me this is not so ...
Michael
Nope, it means specifically xtrans sensor based raw files. Fuji still makes Bayer pattern sensor based cameras (like the X-A3) and these work with OpenCL.
Xtrans uses a special pipeline made specifically for those types of sensor-arrays.
I will encourage you to write to our Support Team with a Feature Request for OpenCL for Xtrans. That way it is easier for us to track demand.0 -
But for the likes of the xt2 or the xpro 2 then opencl is disabled?
This is very disappointing, I will certainly file a support request.
Thanks.0 -
[quote="mtissington" wrote:
But for the likes of the xt2 or the xpro 2 then opencl is disabled?
This is very disappointing, I will certainly file a support request.
Thanks.
It is not disabled, it is simply not implemented at this stage.0 -
I'm interested to understand on what basis you think it is very disappointing.
Right now no one can really know whether it is disappointing or not.
In any case the use of GPUs is not that big an issue for most people in many situations.
If your GPU has enough power it's useful for batch mass processing sessions so long as you have not used some features that are not currently compatible.
For editing it probably means a lot to people working with large number of images - wedding photographers and studio fashion and product shoots for example - but I often take a few thousand shots at events and processing them afterwards on a machine that has very little GPU power and seems to be around the same speed with or without GPU in use, is about as fast as I can manage.
Admittedly I'm not working with the largest files in the world but then when I do download some sample files to engage with, speed is not an issue for me whether OpenCL is active or not.
Thus my question about why you think it to be very disappointing. How would one know?
Grant0 -
The reason for being disappointed is based on have a number of Nikon RAW files from a D810 and now my Fuji XT2.
My machine has 32GB RAM
When I use Adjustment layers and AutoMask there is a huge difference in how long it takes to apply the mask.
With AutoMask disabled there is still a huge difference. Nikon raw files are almost instant.
Adobe Lightroom does not have this difference... but I prefer the image quality from C1.
Michael0 -
Out of interest can you quantify how much of a difference there is?
My notebook workstation has a fairly low power GPU. For a long time it was being ignored for OpenCL purposes.
When it started to be used (according to a GPU monitor program I have installed) I really did not notice the difference.
The RAW files I normally work with are smaller than you will be seeing from you cameras which might explain the perceived lack of slowness.
However when I load up a larger file to experiment with from time to time I still don't find the difference that great. Sometimes just about noticeable but even at 100% longer it would not be very long.
This is now a 4 year old machine running an i7 processor of those times, an SSD (2 actually but one on a slower bus) and 24GB RAM. So nothing really special about it in the performance stakes.
I usually have at least 3 or 4 other applications running - Firefox (with a lot of open tabs), Outlook, Skype (normal and Business) and at least one other application, sometimes more.
There in nothing "tuned" about the setup.
If I run with a session off an external spinning disk via USB3 it it a little slower and the disks sometimes seems to take a pause for a few seconds but that's not a C1 issue as far as I know. It's not usually obvious that I'm using an external drive once it has woken form its "sleep" mode. (With USB3 connections that can sometimes require some persuading. If I use a USB2 socket to wake the drive it responds quickly.)
Maybe using sessions makes a difference?
I'm interested because my "second" machine is now over ten years old and I'm considering a new purchase and shuffling the existing main machine down the list a little. However justifying the "need" to buy nearer the top end of specifications is not so easy. Seeking something will be better than I have now and equally future proof as has this one been has not yet produced a "stand out" option. With notebooks the choice is very limited at the higher end of the GPU options and heading that way adds considerably to the costs with out seeming to add that much in real absolute number crunching capability to the GPU.
Frankly I would not be much concerned .... except for the comment that appear here from time to time!
One conclusion I can firmly acknowledge is that the chance of any new machine being a Mac is zero unless something very dramatic happens!
Grant0 -
[quote="SFA" wrote:
My notebook workstation has a fairly low power GPU. For a long time it was being ignored for OpenCL purposes.
When it started to be used (according to a GPU monitor program I have installed) I really did not notice the difference.
The difference obviously depends on the power of the GPU involved, but really, a low power mobile workstation GPU is not the obvious candidate based on previous comments from Phase One in these forums.
The standard recommendation I've seen here is that for openCL performance, get a gamer GPU not a workstation GPU, and if you can, get AMD not nVidia. My desktop with a (now oldish) AMD GPU does see a nice performance boost from having openCL enabled.0 -
[quote="RobbieAB" wrote:
[quote="SFA" wrote:
My notebook workstation has a fairly low power GPU. For a long time it was being ignored for OpenCL purposes.
When it started to be used (according to a GPU monitor program I have installed) I really did not notice the difference.
The difference obviously depends on the power of the GPU involved, but really, a low power mobile workstation GPU is not the obvious candidate based on previous comments from Phase One in these forums.
The standard recommendation I've seen here is that for openCL performance, get a gamer GPU not a workstation GPU, and if you can, get AMD not nVidia. My desktop with a (now oldish) AMD GPU does see a nice performance boost from having openCL enabled.
Yep, I know what the recommendations are and that my GPU, as I mentioned, offers little real performance and so little real benefit.
However the point is that performance of the current configuration, whichever way it is running (with or without GPU in play), still seems to be good enough that I think I would be challenged to take advantage of any enhanced performance that might become available.
Running an output process with three or four recipes and a couple of thousand files might be a different matter although I am now way feeling things are slow as they are.
Thus I am not entirely sure that expectations will, for many people, be fully satisfied and for many others they probably don't really need whatever may be available to them over and above what they have.
That said there are no doubt users who would benefit a tremendous amount but they are probably relatively few and far between in reality. Most will find a little improvement most of the the time and may not actually be able to tell.
Just my opinion of course.
Grant0 -
If your GPU is barely supported and isn't fast enough to show a difference how can you be critical of others needs?
I desktop machine with a even moderate gaming video card can make an immense difference in editing and especially exporting. On a large event shoot it could save me 10 minutes or more exporting when I was using Sony cameras before moving to Fuji.
A current gaming video card in a desktop is a POWERFUL processing engine. Basic laptop mobile versions cannot even begin to compare.
Personally, I think Phaseone's lack of support for many of these Fuji functions has less to do with the Xtrans sensor and more to do with Fuji GFX medium format that's coming out and will compete with Phase One cameras. Initially Phase One said they would not support the GFX in C1, but quickly came back with "we'll see".
It smells like a pissing match to me more than anything.0 -
[quote="rckstflbn" wrote:
If your GPU is barely supported and isn't fast enough to show a difference how can you be critical of others needs?
I desktop machine with a even moderate gaming video card can make an immense difference in editing and especially exporting. On a large event shoot it could save me 10 minutes or more exporting when I was using Sony cameras before moving to Fuji.
Because my point was and is that many people will likely not notice a difference and absent any comparative results are unlikely to be able to assess one way or another whether it matters to them.
Your 10 minute saving is an interesting assessment but " a big event shoot" is hardly quantifiable and you don't mention things like file size differences.
10 minutes difference could be 100% longer or 10% longer for example. If the shoot is big enough 10 mins may not be a big deal in a typical workflow.
I assume it's not the sort of thing you could easily test for before making your decision to change over or that the time difference reported was not enough of a negative result to stop you changing systems?
As for your subsequent speculations ... possibly against the forum rules but certainly somewhat unnecessary in the context of the original question when posting on a board provided for User to User discussions by the company about whom you are speculating.
Just my opinion of course.
Grant0
投稿コメントは受け付けていません。
コメント
11件のコメント